Chris Oswald
commited on
Commit
•
fbb72cc
1
Parent(s):
38e8e2b
added raw image and mask objects
Browse files
SPIDER.py
CHANGED
@@ -24,6 +24,7 @@ import numpy as np
|
|
24 |
import pandas as pd
|
25 |
|
26 |
import datasets
|
|
|
27 |
import skimage
|
28 |
import SimpleITK as sitk
|
29 |
|
@@ -110,7 +111,7 @@ class CustomBuilderConfig(datasets.BuilderConfig):
|
|
110 |
class SPIDER(datasets.GeneratorBasedBuilder):
|
111 |
"""TODO: Short description of my dataset."""
|
112 |
|
113 |
-
DEFAULT_WRITER_BATCH_SIZE =
|
114 |
|
115 |
VERSION = datasets.Version("1.1.0")
|
116 |
|
@@ -169,9 +170,9 @@ class SPIDER(datasets.GeneratorBasedBuilder):
|
|
169 |
features = datasets.Features({
|
170 |
"patient_id": datasets.Value("string"),
|
171 |
"scan_type": datasets.Value("string"),
|
172 |
-
|
|
|
173 |
"image_array": datasets.Array3D(shape=image_size, dtype='float64'),
|
174 |
-
# "raw_mask": datasets.Image(),
|
175 |
"mask_array": datasets.Array3D(shape=image_size, dtype='float64'),
|
176 |
"metadata": {
|
177 |
"num_vertebrae": datasets.Value(dtype="string"), #TODO: more specific types
|
@@ -478,20 +479,34 @@ class SPIDER(datasets.GeneratorBasedBuilder):
|
|
478 |
image_path = os.path.join(paths_dict['images'], 'images', example)
|
479 |
image = sitk.ReadImage(image_path)
|
480 |
|
|
|
|
|
|
|
481 |
# Convert .mha image to standardized numeric array
|
482 |
-
|
483 |
-
|
|
|
484 |
)
|
485 |
|
|
|
|
|
|
|
486 |
# Load .mha mask file
|
487 |
mask_path = os.path.join(paths_dict['masks'], 'masks', example)
|
488 |
mask = sitk.ReadImage(mask_path)
|
489 |
|
|
|
|
|
|
|
490 |
# Convert .mha mask to standardized numeric array
|
491 |
-
|
492 |
-
|
|
|
493 |
)
|
494 |
-
|
|
|
|
|
|
|
495 |
# Extract overview data corresponding to image
|
496 |
image_overview = overview_dict[scan_id]
|
497 |
|
@@ -502,13 +517,13 @@ class SPIDER(datasets.GeneratorBasedBuilder):
|
|
502 |
return_dict = {
|
503 |
'patient_id':patient_id,
|
504 |
'scan_type':scan_type,
|
505 |
-
'raw_image':
|
506 |
-
'raw_mask':
|
507 |
-
'image_array':
|
508 |
-
'mask_array':
|
509 |
'metadata':image_overview,
|
510 |
'rad_gradings':patient_grades_dict,
|
511 |
-
|
512 |
|
513 |
# Yield example
|
514 |
yield scan_id, return_dict
|
|
|
24 |
import pandas as pd
|
25 |
|
26 |
import datasets
|
27 |
+
import PIL
|
28 |
import skimage
|
29 |
import SimpleITK as sitk
|
30 |
|
|
|
111 |
class SPIDER(datasets.GeneratorBasedBuilder):
|
112 |
"""TODO: Short description of my dataset."""
|
113 |
|
114 |
+
DEFAULT_WRITER_BATCH_SIZE = 16 # PyArrow default is too large for image data
|
115 |
|
116 |
VERSION = datasets.Version("1.1.0")
|
117 |
|
|
|
170 |
features = datasets.Features({
|
171 |
"patient_id": datasets.Value("string"),
|
172 |
"scan_type": datasets.Value("string"),
|
173 |
+
"raw_image": datasets.Image(decode=False),
|
174 |
+
"raw_mask": datasets.Image(decode=False),
|
175 |
"image_array": datasets.Array3D(shape=image_size, dtype='float64'),
|
|
|
176 |
"mask_array": datasets.Array3D(shape=image_size, dtype='float64'),
|
177 |
"metadata": {
|
178 |
"num_vertebrae": datasets.Value(dtype="string"), #TODO: more specific types
|
|
|
479 |
image_path = os.path.join(paths_dict['images'], 'images', example)
|
480 |
image = sitk.ReadImage(image_path)
|
481 |
|
482 |
+
# Convert .mha image to original size numeric array
|
483 |
+
image_array_original = sitk.GetArrayFromImage(image)
|
484 |
+
|
485 |
# Convert .mha image to standardized numeric array
|
486 |
+
image_array_standardized = standardize_3D_image(
|
487 |
+
image_array_original,
|
488 |
+
resize_shape,
|
489 |
)
|
490 |
|
491 |
+
# Create PIL image object of original image
|
492 |
+
PIL_original_image = PIL.Image.fromarray(image_array_original)
|
493 |
+
|
494 |
# Load .mha mask file
|
495 |
mask_path = os.path.join(paths_dict['masks'], 'masks', example)
|
496 |
mask = sitk.ReadImage(mask_path)
|
497 |
|
498 |
+
# Convert .mha mask to original size numeric array
|
499 |
+
mask_array_original = sitk.GetArrayFromImage(mask)
|
500 |
+
|
501 |
# Convert .mha mask to standardized numeric array
|
502 |
+
mask_array_standardized = standardize_3D_image(
|
503 |
+
mask_array_original,
|
504 |
+
resize_shape,
|
505 |
)
|
506 |
+
|
507 |
+
# Create PIL image object of original mask
|
508 |
+
PIL_original_mask = PIL.Image.fromarray(mask_array_original)
|
509 |
+
|
510 |
# Extract overview data corresponding to image
|
511 |
image_overview = overview_dict[scan_id]
|
512 |
|
|
|
517 |
return_dict = {
|
518 |
'patient_id':patient_id,
|
519 |
'scan_type':scan_type,
|
520 |
+
'raw_image':PIL_original_image,
|
521 |
+
'raw_mask':PIL_original_mask,
|
522 |
+
'image_array':image_array_standardized,
|
523 |
+
'mask_array':mask_array_standardized,
|
524 |
'metadata':image_overview,
|
525 |
'rad_gradings':patient_grades_dict,
|
526 |
+
}
|
527 |
|
528 |
# Yield example
|
529 |
yield scan_id, return_dict
|